Bath University Computing Services (BUCS) is planning engineering work from 4:30 pm on Friday 27 March until 9:00 am on Monday 30th March 2009. This means that no UKOLN Web sites or services will be available for that period. Further information is available on the BUCS Web site.
As a variety of UKOLN services will be unavailable over the period (which is the weekend after next) we will need to ensure that our key stakeholders are informed (including our funders, JISC and MLA) and take steps to ensure that we alert anyone who may be making use of such services over this period – and possibly afterwards, if any unexpected problems are encountered.
Before alerting the key stakeholders we needed to identify affected services. As well as the obvious Web sites on a .ukoln.ac.uk domain there are also the Web sites, such as Exploit Interactive (http://www.exploit-lib.org/) and Cultivate Interactive (http://www.cultivate-int.org/) which, although they are hosted locally, do not have an obvious dependency on UKOLN servers.
There was also a need to identify other network services besides Web sites. Being unable to send email messages or receive incoming email may be obvious, but do we have any services which rely on automated processing of emails (such as various Listserv mailing lists we host?) Similarly what about other networked services besides Web and email – what about any LDAP services, streaming video services, Z39.50 services, etc. , etc.? And what about the services outside of Bath which may make use of our services? Will they degrade gracefully if our servers are unavailable over the weekend or mwill such services (which are not only external to us, but we may not even know they exist) fail or timeout as they await a response from our servers?
Having (we hope!) identified the key services we need to disseminate the news of the unavailability of our services and the possible implications for other service providers who have dependencies on our services and the end user communities we need to make use of the various dissemination challenges in order to alert the various affected communities.
Clearly email has an important role to play in communication with the key stakeholders. And we have provided an alert on UKOLN’s news service, which is also available via RSS. These are the obvious dissemination channels, but what else can we use?
In this blog post about the associated issues (which I’ll expand on in the following section) I’m also alerting readers of this blog (who may also be users of UKOLN – and Bath – services) of the scheduled downtime. And I will also use Twitter to send out an announcement about this post which will be followed by another tweet shortly before the services are brought down.
I’ve also updated the RSS feed for the QA Focus Web site and will do something similar shortly for the Exploit Interactive and Cultivate Interactive news feeds.
For this scheduled downtime we have had time to discuss the implications and make plans for informing our users. And we’ve had useful discussions with other affected parties in the University, including the e-learning unit. But what about the wider issues such as whether a weekend of service down-time should be regarded as acceptable, whether we should provide mechanisms for prov9ding backup services which aren’t dependent on the local network or even looking to migrate our services to external providers?
We, of course, aren’t alone in having to consider such issues. Last week there were a number of Twitter posts about service problems with a number of MIMAS services including COPAC and the Archives Hub. And although a MIMAS news item was published when the service was restored I felt that the various tweets which were published when the services first became at risk demonstrated how Twitted can be useful in immediate feedback and also a mechanism for feedback.
Back in January 2008 I wrote a post entitled When Web Sites Go Down which was concerned with the announcement by the University of Southampton that its Web site was down for scheduled maintenance from 2-4th January 2008. In light of the service unavailable of well-established services hosted by prestigious institutions such as the universities of Bath, Manchester and Southampton it might be timely to ask ourselves whether educational institutions need still to be involved in the hosting of widely used services? Wouldn’t it be better, we may ask, to leave hosting to the global organisations such as Google and Yahoo? But if that’s your view, reflect on a recent email sent out by Yahoo to users of the Yahoo Mail service:
Beginning the evening of Friday March 13th (PDT) you may experience problems accessing your Yahoo! Mail account. If your account is affected, it should be available again by midday on Saturday March 14th (PDT).
We sincerely apologize for this inconvenience.
The Yahoo! Mail team
I think we do need to keep asking such questions. But we also need to remember that the grass isn’t always greener on the other side of the fence. And I hope the email send by Yahoo’s support team on the 10 March about the downtime on 13-14 March wasn’t the only notification which Yahoo Mail users received!
But as well as asking ourselves the longer term question about how our services should be hosted, we still need to address the issues of service downtime (whether scheduled or not) and how we alert our users and other service providers who may be affected. Any thoughts?